KR20200065144A - Apparatus and method for automatic inspection of a detailed map using object recognition - Google Patents

Abstract

The present invention relates to an apparatus and a method for automatic inspection of a precision map using object recognition. The apparatus comprises: a camera to photograph the outside of a vehicle; a location measurement unit to measure the location of the vehicle; a precision map database storing a precision map including property information for an object on a road; and a control unit to recognize the object on the road in an image photographed by the camera, calculate a relative location between the recognized object and the vehicle, calculate an absolute location of the recognized object in accordance with location information of the vehicle by the location measurement unit to calculate recognition information of the object, and compare the calculated recognition information of the object with the precision map to determine whether the precision map is erroneous. The recognition information of the object includes the absolute location and type of the object.

Description

Apparatus and method for precision map automatic inspection using object recognition{APPARATUS AND METHOD FOR AUTOMATIC INSPECTION OF A DETAILED MAP USING OBJECT RECOGNITION}

The present invention relates to an apparatus and method for automatically inspecting a precision map using object recognition, and more specifically, compares information of an object recognized through a camera with a precision map stored in the precision map DB, and automatically stores a pre-stored precision map. The present invention relates to an apparatus and method for automatically inspecting a precision map using object recognition to enable inspection.

In general, a navigation device is carried by a user or installed on a vehicle, such as a vehicle, to provide an accurate distance and required time from a current location to a destination, and to guide a route to a destination.

To determine the exact location of a moving object, the current navigation system uses a global navigation satellite system (GNSS) to track the position of a target on the ground using a satellite network.

In such a navigation system, it is essential to use map data for route guidance, and recently, by using a precise map including information on road width, rider, curvature, gradient, and facilities, as well as link and node information on the road. There are increasing cases of providing route guidance.

In addition, with the development of technology, interest in a vehicle capable of autonomous driving is increasing, and in such an autonomous vehicle, a technique of generating a route using a precise map and performing driving is applied.

However, in order to perform autonomous driving using such a precise map, it is necessary to use lanes, road facilities, and sign facility information, so it is important to provide such information with very precise accuracy.

Therefore, it is required to identify and correct the facility, such as a sign, depending on the maintenance of the road, or when there is an error in the precision map initially generated.

Meanwhile, the background technology of the present invention is disclosed in Korean Patent Registration No. 10-0456377 (2004.11.01).

The present invention compares the information of the object recognized by the camera and the precision map stored in the precision map DB, automatically inspects the pre-stored precision map, and performs a precision map automatic inspection device using object recognition to update the precision map. And its purpose.

A precision map automatic inspection device using object recognition according to the present invention includes a camera for photographing the exterior of a vehicle; Position measuring unit for measuring the position of the vehicle; A precision map DB that stores a precision map including property information on objects on the road; And an object on the road from the image captured by the camera, calculates a relative position between the recognized object and the vehicle, and calculates the absolute position of the recognized object according to the location information of the vehicle through the position measuring unit. It includes a control unit for calculating the recognition information of the, and comparing the calculated recognition information of the object with the precision map to determine whether the precision map is an error, the recognition information of the object includes the type and absolute position of the object It is characterized by.

In the present invention, the control unit is characterized by applying an error range according to the positioning method of the position measuring unit when comparing the calculated recognition information of the object with a precision map.

In the present invention, the control unit is characterized in that it compares the recognition information of the calculated object with the precision map in a region around the coordinates of the recognized object and the radius is the error range.

In the present invention, if it is determined that an error exists in the precision map, the controller updates the precision map stored in the precision map DB according to the recognition information of the object.

In the present invention, if it is determined that an error exists in the precision map, the controller transmits recognition information of the object to a server.

In the present invention, the server verifies the accuracy of the received identification information of the object, and when it is determined that the identification information of the object is correct, updates the precision map stored in the server according to the identification information of the object. Is done.

In the present invention, the server is characterized in that when the recognition information of the same object is received from a plurality of vehicles, the recognition information of the object is determined to be correct.

A method for automatically inspecting a precision map using object recognition according to an aspect of the present invention includes: a control unit recognizing an object on a road from an image photographed through a camera; Calculating a relative position between the recognized object and the vehicle by the control unit; Calculating, by the control unit, the absolute position of the recognized object according to the position information of the vehicle and the calculated relative position, and calculating the recognition information of the object; Determining, by the control unit, whether the precision map is in error by comparing the calculated recognition information of the object with a precision map stored in the vehicle; And when it is determined that an error exists in the precision map, the control unit updating the precision map according to the recognition information of the object.

In the present invention, the recognition information of the object is characterized by including the type and absolute position of the object.

In the step of determining whether the precision map of the present invention is in error, the control unit applies an error range according to the positioning method of the vehicle to compare the recognition information of the object with the precision map.

A method for automatically inspecting a precision map using object recognition according to another aspect of the present invention includes: a control unit recognizing an object on a road from an image photographed through a camera; Calculating a relative position between the recognized object and the vehicle by the control unit; Calculating, by the control unit, the absolute position of the recognized object according to the position information of the vehicle and the calculated relative position, and calculating the recognition information of the object; Determining, by the control unit, whether the precision map is in error by comparing the calculated recognition information of the object with a precision map stored in the vehicle; And when it is determined that an error exists in the precision map, the control unit transmitting the recognition information of the object to the server.

A method for automatically inspecting a precision map using object recognition according to another aspect of the present invention includes: verifying the accuracy of the recognition information of the object received by the server; And if the recognition information of the object is determined to be correct, the server further updating the precision map stored in the server according to the recognition information of the object.

In the step of verifying the accuracy of the recognition information of the object of the present invention, the server is characterized in that when the recognition information of the same object is received from a plurality of vehicles, the recognition information of the object is determined to be correct.

The apparatus and method for automatically inspecting a precision map using object recognition according to the present invention can determine whether an error is in the precision map through comparison between the object recognized in the vehicle and the precision map, and automatically update the precision map accordingly. By doing so, there is an effect of ensuring that the latest accurate map is always maintained.

1 is a block diagram showing the configuration of a precision map automatic inspection device using object recognition according to an embodiment of the present invention.
2 is a flowchart illustrating a method for automatically inspecting a precision map using object recognition according to an embodiment of the present invention.
3 is another flowchart for explaining a method for automatically inspecting a precision map using object recognition according to an embodiment of the present invention.

Hereinafter, an embodiment of an apparatus and method for automatically inspecting a precision map using object recognition according to the present invention will be described with reference to the accompanying drawings. In this process, the thickness of the lines or the size of components shown in the drawings may be exaggerated for clarity and convenience. In addition, terms to be described later are terms defined in consideration of functions in the present invention, which may vary according to a user's or operator's intention or practice. Therefore, definitions of these terms should be made based on the contents throughout this specification.

1 is a block diagram showing the configuration of a precision map automatic inspection device using object recognition according to an embodiment of the present invention.

As shown in FIG. 1, the apparatus for automatically inspecting a precision map using object recognition according to an embodiment of the present invention includes a control unit 100, a camera 110, a position measurement unit 120, a sensor unit 130 and precision Map DB 140 is included. In addition, the control unit 100 may communicate with the server 200 through a communication module (not shown), and may be configured to transmit data for updating a precise map to the server 200. The communication method between the control unit 100 and the server 200 may be applied to various methods that have already been utilized.

The apparatus for automatically inspecting a precision map using object recognition may be installed in each vehicle and perform inspection on the precision map stored in the vehicle.

The camera 110 is configured to photograph the outside of the vehicle (vehicle front, side, rear, etc.), and may be provided with a plurality.

The position measuring unit 120 may measure the position of the vehicle. The position measuring unit 120 may include components such as a GNSS system, commonly referred to as a Global Positioning System (GPS), a Dead Reckoning system, a Real-Time Kinematic (RTK) GPS system, and a Differential GPS (DGPS) system. Thereby, it is possible to calculate the position information of the host vehicle.

The sensor unit 130 may be provided with various sensors to grasp the external situation of the vehicle. For example, various sensors may be provided, such as an altitude sensor that measures the altitude of a vehicle, a gyro sensor that measures a vehicle's posture, a rider sensor or a radar sensor that detects distance information between an object and a vehicle outside the vehicle.

On the other hand, in some embodiments of the present invention, it is also possible that a sensor included in the guessing navigation system of the position measuring unit 120 functions as a sensor included in the sensor unit 130.

The precision map DB 140 is composed of a storage device such as a non-volatile memory, and stores the precision map. These precision maps store more information than general maps, and specifically, data corresponding to additional information (property information) about objects on the road, such as lane information, road slope, signs, milestones, traffic lights, and speed cameras. It contains. Here, the attribute information may include the type of the object.

The control unit 100 may recognize objects (lanes, signs, milestones, traffic lights, speed control cameras, etc.) outside the vehicle through the camera 110, and determine the relative positional relationship (distance) between the recognized object and the vehicle. Can be calculated.

For example, in the case of detecting a facility using a camera, it is possible to determine the actual facility size and the relative distance from the vehicle according to the location and size on the facility image, and this technology is already widely used in vehicle image sensing. Is being used.

In addition, even in the case of using a rider or a radar, since it is already a well-known technique to grasp the relative positional relationship between the detected object and the vehicle, the controller 100 can grasp the location information of the facility using various methods.

At this time, since the control unit 100 can grasp the current position of the vehicle through the position measuring unit 120, the absolute position of the recognized object may be calculated according to the current position information of the vehicle and the recognized object and the relative position between the vehicles. .

In addition, the control unit 100 may determine what kind of object (lane, sign, milestone, traffic light, speed control camera, etc.) is an object through processing of the image recognized through the camera.

Accordingly, the control unit 100 may recognize the type and absolute position of the object as recognition information of the object.

The controller 100 may inspect the precision map by comparing the information of the object on which the coordinates are calculated and the property information on the precision map of the coordinate. That is, if the property information of the same object exists at the point corresponding to the location of the object information on the precision map, the control unit 100 determines that the actual road environment and the precision map match, and if the property information of the same object does not exist, the actual It can be judged that the road environment and the precision map do not match.

At this time, the control unit 100 may select an error range of the map inspection in consideration of the error of the positioning system, that is, the position measurement unit 120. In other words, since there is an error in the positioning system of the vehicle, it is necessary to carry out inspection in consideration of such an error to reduce the occurrence of inconsistency determination due to the error. In addition, since the accuracy of the position of the vehicle varies depending on the type of positioning system used for positioning of the vehicle, a variable error range may be applied in consideration of this.

For example, when using a high-precision positioning system such as RTK, the error range is set within 10 cm, and when using a system with the same precision as DGPS, the error range is set within 1 m, and a system with precision such as GPS is used. In the case of, the error range can be set within 5 m.

That is, in the case of using DGPS, if the property information of the same object as the object exists in an area (sphere, circle) with a radius of 1 m centered on the coordinates of the recognized object, it is determined that the object matches, and the object in the area If the property information of the same object as does not exist, it may be determined to be inconsistent.

If it is determined that the actual road environment and the precision map do not match, the controller 100 may update the precision map based on the recognition information of the recognized object. That is, it is possible to store fast-track information of the corresponding object at the location of the recognized object on the precise map.

Alternatively, if it is determined that the actual road environment and the precision map do not match, the control unit 100 transmits the recognition information of the corresponding object to the server 200, so that the information recognized by the server 200 is verified and the precision map is updated. It can be done.

The server 200 may receive recognition information of an object determined to be different from a precision map mounted in each vehicle from a plurality of vehicles, synthesize them, determine whether there is an error in the precision map data, and update the precision map. .

Specifically, the server 200 is based on the recognition information of the received object, when the recognition information for the same object is collected from a plurality of vehicles, the recognition information of the object is correct (that is, there is an error in the precision map) You can judge.

If it is determined that there is an error in the precision map, the server 200 may update the precision map based on the recognition information of the object. That is, the object's location information can be stored on the precision map at the location of the object.

At this time, the server 200 may use a method of determining an average position of a plurality of position values received for the same object as the position of the corresponding object.

That is, the server 200 may verify the accuracy of the received identification information of the object, and when it is determined that the identification information of the object is correct, update the precision map stored in the server according to the identification information of the object.

Meanwhile, the control unit 100 may be configured to periodically perform object recognition and comparison with a precision map.

For example, the control unit 100 may perform the comparison of the property information of the precision map with respect to the alignment perpendicular to the direction of travel at the position of the vehicle with the recognized object information every set interval (eg, 1 m).

That is, such a comparison makes it possible to compare attribute information of the lane alignment at set intervals. At this time, the map linear attribute information may be the number of lanes and the attribute information of each lane.

In addition, when periodically comparing the attribute information of the lane alignment, the control unit 100 may also perform comparison on the corresponding object when additional objects are recognized.

On the other hand, the control unit 100 may use the information measured by the sensor unit 130 to determine whether the property information of the precision map is in error.

For example, by comparing the inclination of the vehicle and the inclination on the precision map of the current location, it is possible to compare whether the property information of the precision map matches the actual environment information. At this time, the control unit 100 may determine whether the match is made with an allowable range in consideration of the error of the sensor.

In addition, in the case of a complicated road such as an overpass or an interchange access road, the control unit 100 is configured to accurately determine which road the vehicle is currently driving using the attitude and altitude information of the vehicle to perform a precision map inspection. Can be.

2 is a flowchart for explaining a method for automatically inspecting a precision map using object recognition according to an embodiment of the present invention, and FIG. 3 is a flowchart illustrating a method for automatically inspecting a precision map using object recognition according to an embodiment of the present invention Is another flow chart.

As shown in FIG. 2, the controller 100 first recognizes an object on the road, and acquires a relative position of the recognized object with respect to the vehicle (S200). For example, in the case of detecting a facility using a camera, it is possible to grasp the actual facility size and a relative distance from the vehicle according to the location and size on the facility image.

Subsequently, the controller 100 calculates the absolute position of the recognized object according to the position information of the vehicle (S210). That is, since the control unit 100 can grasp the current position of the vehicle through the position measuring unit 120, the absolute position of the recognized object may be calculated according to the current position information of the vehicle and the recognized object and the relative position between the vehicles. have.

Thereafter, the controller 100 compares the recognized object and the precision map in consideration of errors according to the positioning method of the vehicle (S220). That is, if the property information of the same object exists at the point corresponding to the location of the object information on the precision map, the control unit 100 determines that the actual road environment and the precision map match, and if the property information of the same object does not exist, the actual It can be judged that the road environment and the precision map do not match.

At this time, since the accuracy of the position of the vehicle is different according to the type of positioning system used for positioning of the vehicle, the controller 100 may apply a variable error range in consideration of this.

According to the comparison of step S220, if there is a difference between the recognized object and the precision map, the controller 100 transmits the recognition information of the object to the server 200 (S230). That is, the automatic map inspection device installed in the vehicle can determine whether or not the actual road environment and the precise map match, and update can be performed on the server, and there is a difference between the actual road environment and the precise map. Only when the object recognition information (location coordinates and object type) is transmitted to the server, the data communication amount can be reduced.

On the other hand, the control unit 100 may transmit the image of the object corresponding to the captured object to the server 200 in addition to the recognition information of the object described above, so that the server 200 can also utilize the image information.

Thereafter, as shown in FIG. 3, the server 200 verifies the recognition information of the received object (S310). For example, when the server 200 collects object identification information from multiple vehicles at the same point based on the received object identification information, the object identification information is correct (that is, an error occurs in the precision map). It can be judged as).

If it is determined that there is an error in the precision map data according to the verification result of the step S320, the server 200 updates the precision map according to the recognition information of the object (S320). That is, the server 200 may update the precision map by storing the speed information of the object in the location of the object on the precision map.

In addition, the server 200 may transmit the updated precise map to the control unit 100 of each vehicle, so that the accurate map update in the vehicle may be performed.

As described above, the apparatus and method for automatically inspecting a precision map using object recognition according to an embodiment of the present invention determine whether an error occurs in the precision map through comparison between the object recognized in the vehicle and the precision map, and update the precision map accordingly. By making it possible to do it automatically, you can always keep up-to-date precision maps.

The present invention has been described with reference to the embodiment shown in the drawings, but this is only exemplary, and those skilled in the art to which the art belongs can various modifications and equivalent other embodiments from this. Will understand. Therefore, the technical protection scope of the present invention should be defined by the following claims.

100: control unit
110: camera
120: position measuring unit
130: sensor unit
140: precision map DB

Claims (14)

A camera that photographs the exterior of the vehicle;
Position measuring unit for measuring the position of the vehicle;
A precision map DB that stores a precision map including property information on objects on the road; And
Recognize the object on the road from the image captured by the camera, calculate the relative position between the recognized object and the vehicle, calculate the absolute position of the recognized object according to the position information of the vehicle through the position measuring unit, Comprising a control unit for calculating the recognition information, and comparing the calculated recognition information of the object with the precision map to determine whether the precision map error,
The recognition information of the object includes the type and absolute position of the object, a precision map automatic inspection device using object recognition.
According to claim 1,
The control unit automatically compares the calculated object recognition information with a precision map, and applies an error range according to the positioning method of the location measurement unit to automatically detect a precision map using object recognition.
According to claim 2,
The control unit, the center of the coordinates of the recognized object and the radius of the error range within the region, the recognition information of the calculated object is compared with the precision map Automatic map recognition device using object recognition, characterized in that.
According to claim 1,
The controller, when it is determined that an error exists in the precision map, updates the precision map stored in the precision map DB according to the recognition information of the object.
According to claim 1,
The controller, if it is determined that an error exists in the precision map, transmits the recognition information of the object to the server, the precision map automatic inspection device using object recognition.
The method of claim 5,
The server verifies the accuracy of the received identification information of the object, and when it is determined that the identification information of the object is correct, an object characterized by updating a precision map stored in the server according to the identification information of the object Precision map automatic inspection device using recognition.
The method of claim 6,
The server, when the recognition information of the same object is received from a plurality of vehicles, the precision map automatic inspection device using object recognition, characterized in that it is determined that the recognition information of the object is correct.
A step in which the control unit recognizes an object on the road from the image photographed through the camera;
Calculating a relative position between the recognized object and the vehicle by the control unit;
Calculating, by the control unit, the absolute position of the recognized object according to the position information of the vehicle and the calculated relative position, and calculating the recognition information of the object;
Determining, by the control unit, whether the precision map is in error by comparing the calculated recognition information of the object with a precision map stored in the vehicle; And
If it is determined that an error exists in the precision map, the controller automatically updates the precision map using object recognition, including the step of updating the precision map according to the recognition information of the object.
The method of claim 8,
A method for automatically inspecting a precision map using object recognition, wherein the object recognition information includes an object type and an absolute position.
The method of claim 8,
In the step of determining whether the precision map is in error, the control unit applies the error range according to the positioning method of the vehicle to compare the recognition information of the object with the precision map, and automatically inspects the precision map using object recognition. Way.
The method of claim 10,
The control unit, based on the coordinates of the recognized object and the radius of the error range in the region, the recognition information of the calculated object is compared with the precision map, characterized in that the automatic detection method of the precision map using object recognition.
A step in which the control unit recognizes an object on the road from the image photographed through the camera;
Calculating a relative position between the recognized object and the vehicle by the control unit;
Calculating, by the control unit, the absolute position of the recognized object according to the position information of the vehicle and the calculated relative position, and calculating the recognition information of the object;
Determining, by the control unit, whether the precision map is in error by comparing the calculated recognition information of the object with a precision map stored in the vehicle; And
If it is determined that an error exists in the precision map, the control unit automatically detects the precision map using object recognition, including the step of transmitting the recognition information of the object to the server.
The method of claim 12,
Verifying the accuracy of the recognition information of the object received by the server; And
When it is determined that the recognition information of the object is correct, the server further updates the precision map stored in the server according to the recognition information of the object. .
The method of claim 13,
In the step of verifying the accuracy of the recognition information of the object, when the server receives the recognition information of the same object from a plurality of vehicles, the precision map automatic using object recognition is characterized in that it is determined that the recognition information of the object is correct Inspection method.

Images